Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Biomed-DPT: Dual Modality Prompt Tuning for Biomedical Vision-Language Models.

Wei Peng, Jianchen Hu, Kang Liu

    IEEE Journal of Biomedical and Health Informatics
    |April 23, 2026
    PubMed
    Summary
    This summary is machine-generated.

    Related Concept Videos

    Improving Translational Accuracy02:07

    Improving Translational Accuracy

    11.5K
    Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
    11.5K
    Improving Translational Accuracy02:07

    Improving Translational Accuracy

    2.6K
    2.6K

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Large-area radiation-modulated thermoelectric fabrics for high-performance thermal management and electricity generation.

    Science advances·2025
    Same author

    Regulating Non-Equilibrium Solvation Structure in Locally Concentrated Ionic Liquid Electrolytes for Wide-Temperature and High-Voltage Lithium Metal Batteries.

    Angewandte Chemie (International ed. in English)·2024
    Same author

    Stretchable photothermal membrane of NIR-II charge-transfer cocrystal for wearable solar thermoelectric power generation.

    Science advances·2023
    Same author

    Effects of nanomaterial exposure on telomere dysfunction, hallmarks of mammalian and zebrafish cell senescence, and zebrafish mortality.

    Ageing research reviews·2023
    Same author

    Inflammatory Genes Associated with Pristine Multi-Walled Carbon Nanotubes-Induced Toxicity in Ocular Cells.

    International journal of nanomedicine·2023
    Same author

    High-Energy-Density Lithium Metal Batteries with Impressive Li<sup>+</sup> Transport Dynamic and Wide-Temperature Performance from -60 to 60 °C.

    Small (Weinheim an der Bergstrasse, Germany)·2023

    Biomed DPT enhances vision-language models for biomedical image classification using dual prompts. This framework improves accuracy on diverse medical datasets, especially for novel classes.

    Area of Science:

    • Biomedical image analysis
    • Artificial intelligence in medicine
    • Computer vision

    Background:

    • Prompt learning is effective for adapting vision-language models (VLMs) to biomedical image classification.
    • Existing methods often overlook crucial visual structures in medical images.
    • A need exists for advanced prompt tuning frameworks that incorporate both textual and visual information.

    Purpose of the Study:

    • To introduce Biomed DPT, a novel knowledge-enhanced dual-modality prompt tuning framework.
    • To improve few-shot learning performance in biomedical image classification.
    • To leverage both clinical and expert-domain knowledge for enhanced VLM adaptation.

    Main Methods:

    • Constructing dual text prompts: ensemble clinical prompts and LLM-driven expert prompts, optimized via a neural network.

    Related Experiment Videos

  • Applying semantic regularization loss to extract clinical knowledge and reduce semantic discrepancies.
  • Utilizing zero vectors as soft visual prompts for attention re-weighting to focus on diagnostic regions.

    • Main Results:

    • Achieved an average classification accuracy of 66.28% across 11 diverse biomedical image datasets (9 modalities, 10 organs).
    • Demonstrated strong performance on base classes (79.54%) and novel classes (76.91%).
    • Effectively focused attention on critical pathological features and diagnostic regions.

    Conclusions:

    • Biomed DPT offers a robust framework for adapting VLMs to biomedical image classification tasks.
    • The dual-modality approach significantly enhances performance in few-shot learning scenarios.
    • The method shows promise for improving diagnostic accuracy and efficiency in medical imaging.